Messages in this thread | | | Date | Sun, 29 Mar 1998 19:49:43 -0500 | From | "Theodore Y. Ts'o" <> | Subject | Re: GGI Project Unhappy On Linux |
| |
Date: Thu, 26 Mar 1998 14:44:44 -0500 From: Vagn Scott <vagn@IDT.NET>
There is a project. The project is GGI. One tiny part of the project is KGI. GGI is big. KGI is small. Only KGI is going into the kernel. GGI is not going into the kernel. Big bad GGI is NOT going into the kernel. Only tiny little bare essential KGI is going in the kernel.
Will putting KGI in the kernel make the kernel useless unless you have GGI installed? The last time I checked, the "small KGI patch" ravaged the existing console code and it wasn't at all clear your system would be usable unless you installed big bad GGI on your system.
KGI is going into the kernel because Xservers crash systems.
My X server doesn't crash. I change back and forth between text mode and the X server all the time and it doesn't crash. My system hasn't crashed due to an X server problem in literally years.
KGI isn't going away because there are no other proposals to deal with the problem that Xservers crash systems.
Fortunately, Linus isn't going to integrate KGI just because it's the only current proposal to solve a problem which is at best academic to most of us. Sure -- from an abstract C.S. point of view a solution like KGI is a good one. However, the last time I looked at the implementation, especially what it did to the existing console code, I was aghast.
The implementation has got to be clean, and ideally, you should be able to integrate KGI --- and then not use GGI if you don't want to use it. The old way of using the existing console driver and existing VT switching code and existing X server should continue to work. My suspicion is that this alone would make people less resistant to integrating KGI.
If you disagree with that, and expect to be taken seriously by GGI/KGI supporters, then propose another solution to the problem that Xservers crash systems.
If you want to be taken seriously, it would help if you lost the attitude. There's a certain amount of fanaticism with the GGI folks which causes them to lose a lot of credibility, at least for some of us. Your attitude really reminds of the BSD folks who have been shrilly insisting on Usenet that asyncronous updates of filesystem metadata is horrible, and so Linux users *must* be losing data all the time. Never mind the fact that people haven't been losing that way --- it still becomes a religious issue for them.
The solution to the problem that X server crash systems? (1) Don't kill -9 them. (2) If you can reliably cause an X server to crash through normal operations (and I can't, even when I'm trying), collect debugging information and send it to the XFree86 folks, and ask them (nicely, instead of with a chip on your shoulder) if they can fix the bug.
- Ted
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |